An integrated decision support system for project risk assessment and treatment decision making considering risk interdependencies

Projects are often embedded in environments characterised by increasing volatility, uncertainty, complexity, and ambiguity. The successful delivery and operation of projects remain critical issues for contemporary project-driven organisations that are exposed to various types of risks. Consequently, an effective project risk management is pivotal to the achievement of project objectives and organisational goals.Project risks are mostly interdependent and can have a cascading impact (propagation effect) on the project objectives. In addition, project risks and their interdependencies can change dynamically throughout a project life cycle (PLC). However, project risk interdependencies are largely ignored or ineffectively addressed in the literature, which can prompt inappropriate project risk assessment (PRA) and reduced efficacy in risk treatment. Furthermore, there is no systematic study that investigates comprehensively the PRA process by incorporating multiple characteristics of project risks and how the influences of different risks on project objectives dynamically evolve over the course of the PLC.This thesis aims to develop an integrated decision support system for improving the decision making processes in PRA and subsequent risk treatment by leveraging suitable analytical and simulation-based methods. The framework of the proposed integrated decision support system consists of three key stages: developing a project risk interdependency network (RIN), assessing project risks based on the RIN, and planning and evaluating project risk treatment actions.Four new PRA solutions focusing on different perspectives of PRA complexity are presented in the integrated decision support system to evaluate project interdependent risks. The first two PRA solutions are devised from the aspect of analytical methods, including Interpretive Structural Modelling (ISM), MICMAC analysis, Social Network Analysis (SNA), and Multi-Criteria Decision Making (MCDM) methods. The remaining two proposed solutions are based principally on simulation-based methods through integrating ISM and Monte Carlo Simulation (MCS).In the first PRA solution developed, a project risk analysis model is proposed by analysing risk interdependencies among project constraints, risks, and objectives throughout the life cycle of a project. A hierarchical project RIN is established using ISM method. The importance of risk/constraint factors associated with project objectives is calculated based on their influence transmission through network paths in the ISM-based RIN. Additionally, MICMAC analysis is employed to complement ISM in investigating both drive and dependence powers of each project RIN element. As a result, critical risk/constraint factors of a project are determined with regard to project objectives, also identifying major project objectives that are highly affected by risk/constraint factors. The green building (GB) project type is selected to demonstrate the proposed risk analysis model. The second PRA solution introduces an MCDM-based PRA framework to comprehensively analyse multiple risk characteristics within a project RIN, which integrates ISM, SNA, and the classical Probability–Impact risk model. An integrated risk indicator, namely risk global prospect is developed using the Entropy TODIM method to prioritise project risks. The indicator edge betweenness centrality is proposed to prioritise risk interdependencies. The overall project risk level is evaluated based on the proposed indicators project local risk loss and project global risk loss. A case study is presented to demonstrate the application of the introduced PRA framework, in which the performances of different risk treatment actions (from a comparative analysis) verify the effectiveness of the proposed Entropy-TODIM based method for PRA compared with other PRA methods. Of the remaining two simulation-based PRA solutions, the third PRA solution develops a new MCS-based RIN model to capture the stochastic behaviour of risk occurrence and to generate numerous risk scenarios when analysing the effects of risk propagation across a project RIN. Then, taking into account the dynamic changes of a project RIN throughout PLC phases, the fourth PRA solution is designed by proposing a prototype of a dynamic MCS-based RIN model, which is able to analyse how the local and global influences of project risks vary with multiple PLC phases. In both simulation-based PRA solutions, five interdependency-based risk indicators are proposed: three indicators to prioritise project risks (the risk's simulated occurrence probability, simulated local influence, and simulated global influence), and two indicators to evaluate the overall project risk level (the project's total risk loss and total risk propagation loss). A sensitivity analysis is conducted to examine the effects of uncertainties of model inputs on PRA results. Moreover, application cases are provided to validate the effectiveness of the proposed MCS-based RIN models for PRA through evaluating and comparing the performances of a series of risk treatment actions.Incorporating a 'network' perspective, this thesis provides significant advances to the application of risk management in an entire PLC through the modelling of complex project risk interdependencies in PRA. The resultant comprehensive and flexible decision support system, enabled by analytical and simulation-based techniques, has great value in practical applications for effective risk assessment and proactive risk treatment in complex projects with large numbers of interdependent risks.